A Study on the Body Temperature Prediction and Safety Diagnosis of Internal Combustion Engine Vehicles Based on Machine Learning

This study designs and implements a machine learning-based temperature prediction and safety diagnosis system for internal combustion engine vehicles. The system features a web-based interface for enhanced user accessibility and employs various algorithms, including Random Forest, Linear Regression, and Support Vector Regression, to improve prediction accuracy. Long Short-Term Memory (LSTM) was also utilized for comparative performance analysis. In the first phase, temperature data from Gwangju Metropolian City, provided by the Korea Meteorological Administration from 2020 to 2024, was extracted for initial experiments, serving as training and validation data for the prediction model. The second phase involved measuring the engine heating temperature of a Hyundai YF Sonata gasoline vehicle over 30 minutes, with the collected data used to evaluate the machine learning model's performance. Additionally, an IoT sensor-integrated web monitoring environment was developed, enabling real-time data collection and providing users with an intuitive interface to enhance temperature prediction. This system offers real-time monitoring capabilities, effectively improving vehicle body safety. The research lays foundational data for exploring the applicability of various vehicle models and is expected to advance body safety management systems.